Data flow: Remove unused column from flowThroughOutOfCall

[hvitved]

I noticed in a run that flowThroughOutOfCall may have a large fan-out when joining call with ccc through matchesCall:

[2024-12-10 20:37:27] Evaluated non-recursive predicate DataFlowImpl::MakeImpl<Location::Location,DataFlowImplSpecific::CsharpDataFlow>::Impl<ExternalLocationSink::LocalFileOutputStreamFlow::C>::Stage2::flowThroughOutOfCall/7#306bfe26@3d1aaa28 in 3916ms (size: 109177854).
Evaluated relational algebra for predicate DataFlowImpl::MakeImpl<Location::Location,DataFlowImplSpecific::CsharpDataFlow>::Impl<ExternalLocationSink::LocalFileOutputStreamFlow::C>::Stage2::flowThroughOutOfCall/7#306bfe26@3d1aaa28 with tuple counts:
           153181   ~8%    {2} r1 = JOIN `DataFlowImplCommon::DataFlowCallEx.projectToCall/0#dispred#c63e421f` WITH `DataFlowImpl::MakeImpl<Location::Location,DataFlowImplSpecific::CsharpDataFlow>::Impl<ExternalLocationSink::LocalFileOutputStreamFlow::C>::Stage1::callMayFlowThroughRev/1#05c410c1` ON FIRST 1 OUTPUT Lhs.0, Lhs.1
           291231   ~5%    {8}    | JOIN WITH `project#DataFlowImpl::MakeImpl<Location::Location,DataFlowImplSpecific::CsharpDataFlow>::Impl<ExternalLocationSink::LocalFileOutputStreamFlow::C>::Stage1::callEdgeReturn/7#10c89f93` ON FIRST 1 OUTPUT Rhs.1, Rhs.5, Rhs.5, Rhs.2, Lhs.0, Lhs.1, Rhs.3, Rhs.4
           289311   ~0%    {7}    | JOIN WITH `DataFlowImpl::MakeImpl<Location::Location,DataFlowImplSpecific::CsharpDataFlow>::Impl<ExternalLocationSink::LocalFileOutputStreamFlow::C>::Stage1::returnMayFlowThrough/4#6a6483e2` ON FIRST 4 OUTPUT Lhs.5, Lhs.4, Lhs.0, Lhs.6, Lhs.7, Lhs.1, Lhs.1
        108888543   ~1%    {7}    | JOIN WITH `project#DataFlowImplCommon::Cached::CachedCallContextSensitivity::getSpecificCallContextCall/2#2f40f4a2` ON FIRST 1 OUTPUT Lhs.1, Rhs.1, Lhs.2, Lhs.3, Lhs.4, Lhs.5, Lhs.6
                       
          1362219   ~3%    {1} r2 = SCAN `DataFlowImplCommon::DataFlowCallEx.projectToCall/0#dispred#c63e421f` OUTPUT In.0
          1362219   ~3%    {1}    | STREAM DEDUP
           153181   ~1%    {1}    | JOIN WITH `DataFlowImpl::MakeImpl<Location::Location,DataFlowImplSpecific::CsharpDataFlow>::Impl<ExternalLocationSink::LocalFileOutputStreamFlow::C>::Stage1::callMayFlowThroughRev/1#05c410c1` ON FIRST 1 OUTPUT Lhs.0
           291231   ~2%    {7}    | JOIN WITH `project#DataFlowImpl::MakeImpl<Location::Location,DataFlowImplSpecific::CsharpDataFlow>::Impl<ExternalLocationSink::LocalFileOutputStreamFlow::C>::Stage1::callEdgeReturn/7#10c89f93` ON FIRST 1 OUTPUT Rhs.1, Rhs.5, Rhs.5, Rhs.2, Lhs.0, Rhs.3, Rhs.4
           289311   ~1%    {6}    | JOIN WITH `DataFlowImpl::MakeImpl<Location::Location,DataFlowImplSpecific::CsharpDataFlow>::Impl<ExternalLocationSink::LocalFileOutputStreamFlow::C>::Stage1::returnMayFlowThrough/4#6a6483e2` ON FIRST 4 OUTPUT Lhs.4, Lhs.0, Lhs.5, Lhs.6, Lhs.1, Lhs.1
           289311   ~4%    {7}    | JOIN WITH `num#DataFlowImplCommon::CallContextSensitivity<Cached::CachedCallContextSensitivity::CallContextSensitivityInput>::TSomeCall#b6cffc7d` CARTESIAN PRODUCT OUTPUT Lhs.0, Rhs.0, Lhs.1, Lhs.2, Lhs.3, Lhs.4, Lhs.5
                       
        109177854   ~1%    {7} r3 = r1 UNION r2
                           return r3

I turns out that the column is actually not needed, since all predicates that join against flowThroughOutOfCall already have both call and ccc, so those should already match.

DCA shows a nice speedup for the C# project mono.

Copilot wasn't able to review any files in this pull request.

Files not reviewed (1)

• shared/dataflow/codeql/dataflow/internal/DataFlowImpl.qll: Language not supported

Tip: Turn on automatic Copilot reviews for this repository to get quick feedback on every pull request. Learn more

[aibaars]

Looks reasonable to me.

[aschackmull]

LGTM